[roland/nptl-sparc] Move SPARC code out of nptl/sysdeps/sparc/.

  Not tested at all, but this should not change anything.
Note it presumes the previously-posted patches on the same branch.


Thanks,
Roland


	* nptl/sysdeps/sparc/Makefile: Moved ...
	* sysdeps/sparc/nptl/Makefile: ... here.
	* nptl/sysdeps/sparc/tcb-offsets.sym: Moved ...
	* sysdeps/sparc/nptl/tcb-offsets.sym: ... here.
	* nptl/sysdeps/sparc/tls.h: Moved ...
	* sysdeps/sparc/nptl/tls.h: ... here.
	* nptl/sysdeps/sparc/sparc32/pthread_spin_lock.S: Moved ...
	* sysdeps/sparc/sparc32/nptl/pthread_spin_lock.S: ... here.
	* nptl/sysdeps/sparc/sparc32/pthread_spin_trylock.S: Moved ...
	* sysdeps/sparc/sparc32/nptl/pthread_spin_trylock.S: ... here.
	* nptl/sysdeps/sparc/sparc32/pthreaddef.h: Moved ...
	* sysdeps/sparc/sparc32/nptl/pthreaddef.h: ... here.
	* nptl/sysdeps/sparc/sparc64/pthread_spin_init.c: Moved ...
	* sysdeps/sparc/sparc64/nptl/pthread_spin_init.c: ... here.
	* nptl/sysdeps/sparc/sparc64/pthread_spin_lock.S: Moved ...
	* sysdeps/sparc/sparc64/nptl/pthread_spin_lock.S: ... here.
	* nptl/sysdeps/sparc/sparc64/pthread_spin_trylock.S: Moved ...
	* sysdeps/sparc/sparc64/nptl/pthread_spin_trylock.S: ... here.
	* nptl/sysdeps/sparc/sparc64/pthread_spin_unlock.S: Moved ...
	* sysdeps/sparc/sparc64/nptl/pthread_spin_unlock.S: ... here.
	* nptl/sysdeps/sparc/sparc64/pthreaddef.h: Moved ...
	* sysdeps/sparc/sparc64/nptl/pthreaddef.h: ... here.
	* nptl/sysdeps/sparc/sparc32/sparcv9/pthread_spin_init.c: Moved ...
	* sysdeps/sparc/sparc32/sparcv9/nptl/pthread_spin_init.c: ... here.
	Update #include.
	* nptl/sysdeps/sparc/sparc32/sparcv9/pthread_spin_lock.S: Moved ...
	* sysdeps/sparc/sparc32/sparcv9/nptl/pthread_spin_lock.S: ... here.
	Update #include.
	* nptl/sysdeps/sparc/sparc32/sparcv9/pthread_spin_trylock.S: Moved ...
	* sysdeps/sparc/sparc32/sparcv9/nptl/pthread_spin_trylock.S: ... here.
	Update #include.
	* nptl/sysdeps/sparc/sparc32/sparcv9/pthread_spin_unlock.S: Moved ...
	* sysdeps/sparc/sparc32/sparcv9/nptl/pthread_spin_unlock.S: ... here.
	Update #include.
  

From: Roland McGrath <roland@hack.frob.com>
Date: Thu, 12 Jun 2014 14:36:33 -0700 (PDT)

> Not tested at all, but this should not change anything.
> Note it presumes the previously-posted patches on the same branch.

Sorry for not reviewing this earlier, I planned to do it earlier today.

It looks fine to me, and if there should be some fallout I'll fix it up
for you.
  

> Sorry for not reviewing this earlier, I planned to do it earlier today.
> 
> It looks fine to me, and if there should be some fallout I'll fix it up
> for you.

Thanks.  I've merged the branch, getting these four commits:

	febf6cc Move SPARC public headers out of nptl/
	e268224 Move SPARC code out of nptl/sysdeps/sparc/.
	b0725f4 SPARC: Define TLS_DEFINE_INIT_TP
	b14368b SPARC: Convert fork.c to arch-fork.h

What would be really great is if you could RSN do the clone and vfork
consolidation as most other machines have done now.  That should leave only
the simple renames left to entirely get rid of
nptl/sysdeps/unix/sysv/linux/sparc/.


Thanks,
Roland
  

From: Roland McGrath <roland@hack.frob.com>
Date: Thu, 12 Jun 2014 15:32:22 -0700 (PDT)

>> Sorry for not reviewing this earlier, I planned to do it earlier today.
>> 
>> It looks fine to me, and if there should be some fallout I'll fix it up
>> for you.
> 
> Thanks.  I've merged the branch, getting these four commits:
> 
> 	febf6cc Move SPARC public headers out of nptl/
> 	e268224 Move SPARC code out of nptl/sysdeps/sparc/.
> 	b0725f4 SPARC: Define TLS_DEFINE_INIT_TP
> 	b14368b SPARC: Convert fork.c to arch-fork.h
> 
> What would be really great is if you could RSN do the clone and vfork
> consolidation as most other machines have done now.  That should leave only
> the simple renames left to entirely get rid of
> nptl/sysdeps/unix/sysv/linux/sparc/.

Ok, I just have to deal with this link failure first:

/home/davem/src/GIT/GLIBC/build-sparcv9/libc_pic.os: In function `fork_handler_alloc':
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:71: undefined reference to `__fork_handlers'
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:71: undefined reference to `__fork_handlers'
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:71: undefined reference to `__fork_handlers'
/home/davem/src/GIT/GLIBC/build-sparcv9/libc_pic.os: In function `__linkin_atfork':
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:114: undefined reference to `__fork_handlers'
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:114: undefined reference to `__fork_handlers'
/home/davem/src/GIT/GLIBC/build-sparcv9/libc_pic.os:/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:114: more un\
defined references to `__fork_handlers' follow
/home/davem/src/GIT/GLIBC/build-sparcv9/libc_pic.os: In function `__libc_pthread_init':
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/libc_pthread_init.c:44: undefined reference to `__fork_generation_pointer'
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/libc_pthread_init.c:44: undefined reference to `__fork_generation_pointer'
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/libc_pthread_init.c:44: undefined reference to `__fork_generation_pointer'
/home/davem/src/GIT/GLIBC/build-sparcv9/libc_pic.os: In function `free_mem':
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:128: undefined reference to `__fork_handlers'
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:128: undefined reference to `__fork_handlers'
/home/davem/src/GIT/GLIBC/glibc/nptl/../nptl/sysdeps/unix/sysv/linux/register-atfork.c:128: undefined reference to `__fork_handlers'
/usr/bin/ld: /home/davem/src/GIT/GLIBC/build-sparcv9/libc.so: hidden symbol `__fork_generation_pointer' isn't defined
/usr/bin/ld: final link failed: Bad value
collect2: ld returned 1 exit status
  

From: David Miller <davem@davemloft.net>
Date: Thu, 12 Jun 2014 15:58:20 -0700 (PDT)

> From: Roland McGrath <roland@hack.frob.com>
> Date: Thu, 12 Jun 2014 15:32:22 -0700 (PDT)
> 
>>> Sorry for not reviewing this earlier, I planned to do it earlier today.
>>> 
>>> It looks fine to me, and if there should be some fallout I'll fix it up
>>> for you.
>> 
>> Thanks.  I've merged the branch, getting these four commits:
>> 
>> 	febf6cc Move SPARC public headers out of nptl/
>> 	e268224 Move SPARC code out of nptl/sysdeps/sparc/.
>> 	b0725f4 SPARC: Define TLS_DEFINE_INIT_TP
>> 	b14368b SPARC: Convert fork.c to arch-fork.h
>> 
>> What would be really great is if you could RSN do the clone and vfork
>> consolidation as most other machines have done now.  That should leave only
>> the simple renames left to entirely get rid of
>> nptl/sysdeps/unix/sysv/linux/sparc/.
> 
> Ok, I just have to deal with this link failure first:

It seems that removing sysdeps/unix/sysv/linux/sparc/fork.S fixes
the problem, doing full 32-bit and 64-bit test builds now.
  

On Thu, 12 Jun 2014, Roland McGrath wrote:

> What would be really great is if you could RSN do the clone and vfork
> consolidation as most other machines have done now.  That should leave only

If there are particular changes in this series you'd like machine 
maintainers to do rather than planning to do yourself, please put a 
description of them (with links to any relevant further details) on the 
PortStatus wiki page, listing the machines for which it is applicable.  
My presumption is that if you haven't put something there but have done 
the change for some machines, you intend to do it for other machines as 
well yourself (and that this applies generally when anyone is making such 
cross-machine changes for which it's predictable what machines need 
updates - the person making the changes will either do them for all 
machines, or add details to PortStatus).
  

> It seems that removing sysdeps/unix/sysv/linux/sparc/fork.S fixes
> the problem, doing full 32-bit and 64-bit test builds now.

Oh, yeah.  That is something that would have been handled by doing the fork
consolidation first.
  

I had planned to at least make the attempt to do all those changes myself
just so as not to make too much work for people and then harangue them
about doing it for me.  Some machine maintainers have beat me to the punch,
which is great.  Dave appeared somewhat eager to help, so I thought he
might take it on.
  

From: Roland McGrath <roland@hack.frob.com>
Date: Thu, 12 Jun 2014 16:52:06 -0700 (PDT)

>> It seems that removing sysdeps/unix/sysv/linux/sparc/fork.S fixes
>> the problem, doing full 32-bit and 64-bit test builds now.
> 
> Oh, yeah.  That is something that would have been handled by doing the fork
> consolidation first.

I think I have the consolidation done properly now too, will commit after
build/check runs complete.
  

Looks great!

Thanks,
Roland